Key points are not available for this paper at this time.
OBJECTIVE: An intelligent agent, Fitting to Outcomes eXpert, was developed to optimize and automate Cochlear implant (CI) programming. The current article describes the rationale, development, and features of this tool. BACKGROUND: Cochlear implant fitting is a time-consuming procedure to define the value of a subset of the available electric parameters based primarily on behavioral responses. It is comfort-driven with high intraindividual and interindividual variability both with respect to the patient and to the clinician. Its validity in terms of process control can be questioned. Good clinical practice would require an outcome-driven approach. An intelligent agent may help solve the complexity of addressing more electric parameters based on a range of outcome measures. METHODS: A software application was developed that consists of deterministic rules that analyze the map settings in the processor together with psychoacoustic test results (audiogram, A(section sign)E phoneme discrimination, A(section sign)E loudness scaling, speech audiogram) obtained with that map. The rules were based on the daily clinical practice and the expertise of the CI programmers. The data transfer to and from this agent is either manual or through seamless digital communication with the CI fitting database and the psychoacoustic test suite. It recommends and executes modifications to the map settings to improve the outcome. RESULTS: Fitting to Outcomes eXpert is an operational intelligent agent, the principles of which are described. Its development and modes of operation are outlined, and a case example is given. Fitting to Outcomes eXpert is in use for more than a year now and seems to be capable to improve the measured outcome. CONCLUSION: It is argued that this novel tool allows a systematic approach focusing on outcome, reducing the fitting time, and improving the quality of fitting. It introduces principles of artificial intelligence in the process of CI fitting.
Govaerts et al. (2010) studied this question.